The current study investigates if and how listeners from tonal and non-tonal varieties of Korean (Kyungsang and Seoul) use different cue weighting in their perception of the three-way distinction among Korean voiceless stops (fortis, lenis, and aspirated). Forty-two Korean listeners (21 each from Seoul and Kyungsang) were tested with stimuli in which VOT and F0 were systematically manipulated. Analyses of the perceptual identification functions show that VOT and F0 cues trade off each other for the perception of the three stops. However, the trading relationship differs between the two dialects. Logistic regression analysis reported interactions among VOT, F0 and Dialect factors for perceiving one stop over the others, indicating the two dialects use the perceptual cues differently. Specifically, while Seoul listeners rely primarily on F0 for making lenis responses and on VOT and F0 for aspirated responses, F0 plays a less important role in modulating both lenis and aspirated responses for Kyungsang speakers. A similar pattern has been observed in production [Lee and Jongman 2012, JIPA 42 (2)]. The results will be discussed in terms of the close link between perception and production across the two different dialects.